ABSTRACT
This chapter introduces three non-standard image coding techniques: vector quantization (VQ), fractal coding, and model-based coding. VQ is an effective technique for performing data compression. Theoretically, VQ is always better than scalar quantization because it fully exploits the correlation between components within the vector. For VQ, block size for code vector and input vector is the same while in partitioned iterated function systems fractal coding the size of the domain block is different from the size of the range blocks. The basic idea of model-based coding is to reconstruct an image with a set of model parameters. Thus, the design decisions in implementing image VQ include vector formation, training set generation, codebook generation, and quantization. In vector formation, it is well known that the image data in the spatial domain can be converted to a different domain so that subsequent quantization and joint entropy encoding can be more efficient.
